R7.1 roadmap, and additional goals

[Philip Prindeville]

Luc Verhaegen wrote:

>On Thu, Apr 13, 2006 at 05:43:48PM -0600, Philip Prindeville wrote:
>  
>
>>Is it too late to be adding additional goals to the R7.1 roadmap?
>>
>>http://xorg.freedesktop.org/wiki/ChangesForX11R71
>>
>>I was thinking that it might be nice to draft an API for supporting
>>various video back-ends for supporting composite, VGA, DVI,
>>LVDS, DFP, etc. and having a standard API for transmitters/encoders/
>>whatnot that's decoupled from the video engine itself, since one often
>>finds board manufacturers (or video controller manufacturers) mixing
>>and matching vendors when it comes to back-ends...  You might have
>>a Cirrus video controller, but it will be using a VIA or Sigma or TI
>>encoder or transmitter... etc.
>>
>>For instance, if support gets hacked into the unichrome drivers for the
>>VT1631 LVDS transmitter, should all of the work have to be duplicated
>>to get the Intel 915 or S3 or Trident drivers to support the same
>>transmitter?  Seems like wasted effort.
>>
>>Or is this more than can be realistically done?
>>
>>-Philip
>>    
>>
>
>As with everything modesetting, and this is a big part of the whole 
>modesetting problem, this needs to grow from the drivers up, not the 
>other way around.
>
>So drafting an API is not very useful. Moulding drivers is.
>
>Luc Verhaegen.
>  
>

Not sure I follow. So how would you do things if you had your way?

Find the best implementation (or most complete partial implementation)
in one
of the drivers, try to separate it out from the driver into its own
module... and
then add similar hooks to other drivers to be able to make use of the same
module?

Have the existing code establish a defacto API?

At least with an API, it gives you the form in broad strokes that the driver
needs to be moulded to, right?

-Philip